Hello, Am 21.12.2007 um 18:23 schrieb Blanco, Jose:
> A user has noticed that while attempting to retrieve a bitstream from > our Dspace instance their firewall blocked the returned information > due > to our system sending a duplicate "Content-length" in the http header. > ... > Via a perl script that makes a request for bitstreams using LWP I > confirmed that there were two Content-length's being returned. Yes, there should not be two headers of the same type in one response. It might be better to have only one, even if he specifies a wrong value or have no content- length header at all. Question is which part of the stack adds the header at which time. As you have this perl script in place, a try and error approach might be suitable. Check against other DSpace instances until you find one that does not create duplicate headers, then check with the owner and compare details of you http server and JSP container (apache + tomcat usually) configuration until you find the difference responsible for this to happen. According to the http protocol, headers have to be sent before the content starts. With an app server the final size of content might not be known at that time. The server could be configured to produce the whole answer, find out about size, adjust headers and send the whole response from cache. This is usually too expensive in terms of memory and response time and is generally not done. It is certainly wrong to correct the header during the response when the first guess was wrong. I guess that e.g. apache is configured to add a header while dspace produces his own header and adds it to the answer before sending content. Verbose logging, debug switches and all the common means are probably required to find out which part of the whole stack adds the superfluous header. Good luck, Christian ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ DSpace-tech mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dspace-tech
